Erschienen in: Science ; 362 (2018), 6413. - S. 457-460 https://dx.doi.org/10.1126/science.aao6809 EVOLUTION qPCRs on F2 Pnye/Hsau hybrid individuals con- firmed that expression differences are linked to the agrp2 locus and exhibit an allelic dosage effect as expected for cis-regulatory mutations Agouti-related peptide 2 facilitates (fig. S4). To identify causal mutations affecting both convergent evolution of stripe agrp2 expression and stripe phenotype, we sequenced the agrp2 locus in individuals (n ≥ patterns across cichlid fish radiations 10 individuals per population) from natural populations of the three hybrid-cross spe- cies. We screened for alternatively fixed, fully Claudius F. Kratochwil1,2,3*, Yipeng Liang1, Jan Gerwin1,3, Joost M. Woltering1, associated variants with the stripe phenotype in Sabine Urban1,3, Frederico Henning1,4, Gonzalo Machado-Schiaffino1,5, pairwise comparisons of each striped species C. Darrin Hulsey1,3, Axel Meyer1,3* (Hsau or Hchi) versus the nonstriped Pnye. Our analyses indicated that a 1.1-kb interval within The color patterns of African cichlid fishes provide notable examples of phenotypic the first agrp2 intron (Fig. 2B) that exhibited convergence. Across the more than 1200 East African rift lake species, melanic horizontal shared, alternatively fixed alleles is a strong stripes have evolved numerous times. We discovered that regulatory changes of the gene candidate region for a regulatory element con- agouti-related peptide 2 (agrp2) act as molecular switches controlling this evolutionarily trolling agrp2 expression (fig. S7). To test whether labile phenotype. Reduced agrp2 expression is convergently associated with the presence this 1.1-kb interval [enhancer of agrp2 in Pnye of stripe patterns across species flocks. However, cis-regulatory mutations are not (Pnye.enh.a)] contains cis-regulatory elements predictive of stripes across radiations, suggesting independent regulatory mechanisms. that could influence interspecific differences be- Genetic mapping confirms the link between the agrp2 locus and stripe patterns. The crucial tween striped and nonstriped species, we tested role of agrp2 is further supported by a CRISPR-Cas9 knockout that reconstitutes stripes theelementsofbothspeciesinagreenfluores- in a nonstriped cichlid. Thus, we unveil how a single gene affects the convergent evolution cent protein (GFP) reporter assay in vivo (sup- of a complex color pattern. plementary text). It showed that Pnye.enh.a efficiently modulates GFP expression [Tukey’s tephen Jay Gould famously posited that if (Pnye, nonstriped, Fig. 1E) and Haplochromis honest significant difference (HSD), P < 0.001] it were possible to rerun the “tape of life,” sauvagei (Hsau,striped,Fig.1C)—we found that and is significantly more potent than the ho- outcomes would be different (1). The rel- horizontal stripes (Fig. 2C) are inherited as a mologous sequence of the striped species Hsau S ativeimportanceofdeterminismandcon- recessive Mendelian trait mapping to chromo- (Tukey’sHSD,P < 0.001) (Fig. 2, E and F, and tingency during evolution is still far from some18(Fig.2A).Thiswasconfirmedbya fig. S8). Together, these results indicate that settled (2, 3). But for particular groups of or- second cross involving the same nonstriped higher expression of agrp2,andtherebythe ganisms, one can now test Gould’s hypothesis. species and another striped species, H. chilotes suppression of stripe patterns, is indeed en- For instance, in less than 8 million to 12 million (Hchi, striped) (Fig. 2A and supplementary text). hanced by Pnye.enh.a (fig. S9). years, more than 1200 species of cichlid fishes To more precisely isolate the causal genetic Our results reveal agrp2 as a major determi- have evolved to form repeated adaptive radia- interval for stripe presence, we fine-mapped the nant of stripe presence that might be sufficient tionsintheEastAfricanRiftValleylakes, trait using recombinant F2 individuals of the to suppress stripe patterns in Pnye.Tofurther such as Lakes Victoria, Tanganyika, and Malawi Pnye × Hsau cross and reduced the causal in- test this finding, we used CRISPR-Cas9 genome (Fig. 1B) (4–8). These adaptive radiations have terval from 600 to 25 kb (fig. S1). This interval engineering to manipulate agrp2 and to thereby given rise to a large diversity of species display- contained the genes agouti-related peptide 2 (agrp2), potentially derepress stripe patterns. Pnye eggs ing various color patterns (Fig. 1, C to N), in- v-type proton ATPase subunit d 2 (atp6V0d2), were injected with Cas9 and agrp2 guide RNAs, cluding the repeated occurrence of melanic and an unknown gene (unk) (Fig. 2B). The re- and we obtained four mutants, all of which had horizontal stripes (Fig. 1A and supplementary sequencing of all coding regions revealed no nonsense and frameshift mutations within agrp2 text). Convergent evolution is prevalent in the fixed missense or nonsense mutations (fig. S2), (fig. S10 and table S1). These CRISPR-Cas9 mu- East African cichlid radiations (9–11), providing suggesting that cis-regulatory variation determines tants developed a continuous midlateral stripe a replicated natural experiment whereby dis- stripe presence. (Fig. 2H and fig. S10) yet no dorsolateral stripe tantly related species from independent adapt- The teleost-specific agrp2 (fig. S3) is a strong (supplementary text). Because horizontal stripes ive radiations can be used to determine what candidate gene for stripes because its paralogs were never observed in noninjected Pnye indi- mechanisms have generated these recurrent have been previously associated with pigmen- viduals (>100 observations; Fig. 2G), this strongly phenotypes (12–16). More specifically, we address tation phenotypes (18–20). To test for agrp2 ex- suggests that although species such as Pnye have whether horizontal stripes, a convergent pheno- pression differences between nonstriped (Pnye) no stripes, the genomic and developmental ma- type, have an identical, similar, or different mo- and striped (Hsau) Lake Victoria cichlids, we per- chinery for stripe pattern formation is in place, lecular bases across the independent adaptive formed quantitative polymerase chain reaction and stripes can reappear in this nonstriped spe- radiations of cichlid fishes. (qPCR; Fig. 2D and fig. S5) on a number of adult cies by experimental manipulation of agrp2. Previously (17)—using a genetic mapping panel tissues, including skin (supplementary text). Here, Next, we tested if the expression levels of of two Lake Victoria species, Pundamilia nyererei agrp2 showed a significantly higher expression in agrp2 and stripe patterns are generally asso- the skin of Pnye (Fig. 2D and fig. S4). The ciated across other cichlid species from the 1Department of Biology, University of Konstanz, Konstanz, lack of consistent expression variation between repeated species flocks of Lakes Victoria, Malawi, Germany. 2Zukunftskolleg, University of Konstanz, Konstanz, melanic and nonmelanic regions and generally and Tanganyika, suggesting a shared molecular 3 Germany. International Max Planck Research School for across dorsoventral and anterior-posterior po- basis for convergent stripe phenotypes. Using Organismal Biology (IMPRS-OB), Max Planck Institute for Ornithology, Konstanz, Germany. 4Department of Genetics, sitions suggests that agrp2 does not shape qPCR on adult skins of striped and nonstriped Institute of Biology, Federal University of Rio de Janeiro pigmentation patterns through local expression- species of each of the three major East African (UFRJ), Rio de Janeiro, Brazil. 5Department of Functional level variation but rather acts as a general stripe cichlid radiations (in total, 24 species; fig. S11), Biology, Area of Genetics, University of Oviedo, Oviedo, pattern inhibitor (fig. S6). Whereas qPCR re- we revealed that nonstriped species commonly Spain. *Corresponding author. Email: claudius.kratochwil@ vealed no such expression differences for paral- had higher agrp2 expression levels than striped uni-konstanz.de (C.F.K.); [email protected] (A.M.) ogs and neighboring genes (supplementary text), species (Fig. 3B). This association was confirmed 1of4 Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-7g8ic3jxx8d7 by comparative phylogenetic analyses (Fig. 3A), A { Lake Victoria radiation ca. 500 species demonstrating a significant evolutionary associ- (horizontal) 0.01 - 1 million years old ation between low agrp2 expression and stripe C stripedE non-striped presence [phylogenetic analysis of variance B (ANOVA); mean P < 0.001; supplementary text]. To determine if this convergence at the phe- Hsau Pnye (vertical) D F notypic and agrp2 gene expression level is also paralleled at the sequence level (16), we com- Hchi Hlat paratively analyzed homologous enh.a sequences across cichlids from Lakes Victoria, Malawi, and Lake Tanganyika radiations Lake Malawi radiation Tanganyika. A tree of enh.a revealed substan- ca. 250 species ca. 500-800 species tial sequence variation and resolved striped 8 - 12 million years old 2 - 4 million years old Lake Victoria species as monophyletic, sug- G striped I non-striped K striped M non-striped gesting a single origin of the striped alleles, whereas striped species of other lakes were Tvit Ncyl Pcya Pdem not monophyletic (Fig. 3C). None of the nine H J L N mutations within enh.a that showed complete association with stripes in Lake Victoria cichlids Jorn Nbri Maur Lcae showed similar stripe association in cichlids of Lakes Malawi